Multi-output regression

[Zhangyh1432]

Hi TabPFN Team,

Thank you for your excellent work on TabPFN! I would like to ask from which version the TabPFNRegressor was introduced. I couldn't find a clear indication in the release notes or documentation.

It would be great if you could clarify this. Thanks in advance for your help!

Best regards

[Zhangyh1432]

I would also like to confirm if the latest version of TabPFN supports multi-output regression. If not, are there any plans to add this feature in the future?

[noahho]

Hi! It does not currently do so, but we will support this soon (The code is mostly ready in tabpfn-extensions repository). Are you looking for a point prediction or distribution for multi-output?
For your first question: It was release in version 2.0

[stlaoweng]

Hello, I have applied the TabPFN in my own datasets with the best performance compared to other ML models.
I am also looking forwards to the multi-output regression ( 1*n output ) , please inform me when the new version is available.

[ZiqiGuo98]

Hi, I am wondering if the multi-output regression is available now in the latest version of tabpfn. Thanks!

[noahho]

This is just being built here: PriorLabs/tabpfn-extensions#207 and will be merged soon. You can already use this by installing from that branch

[klemens-floege]

@noahho what is the state on this? I see that PriorLabs/tabpfn-extensions#207 was never merged; shall we leave the Issue open, merge the fix or add a "help wanted" tag?